IJCS | Volume 33, Nº4, July and August 2020

308 Pereira et al. Patent foramen ovale in cryptogenic stroke Int J Cardiovasc Sci. 2020; 33(4):307-317 Original Article no room for excessive optimism. These relatively modest results have been attributed to the choice of closure device, off-protocol closure device use within the medical therapy arms, patient selection criteria and slow enrolment, 1,3,5,7,8 among other reasons. In the years 2017 and 2018, three new clinical trials were published, which demonstrated that percutaneous PFO closure, as compared with medical therapy, does reduce the risk of recurrent stroke. 6,9,10 Some of these results, impressive as they are, have been obtained by the selective inclusion of patients with high-risk PFO features, including the size of the patent foramen ovale, or the presence of an atrial septal aneurysm, making PFO closure particularly persuasive in these patients. However, restricting device closure entirely to patients with high-risk characteristics of PFO may be too conservative. 11 Concerning the clinical trials currently published, several meta-analyses were carried out, 12-19 but all of them include data from a follow-up study 2 rather than the original clinical trial data, that is, data that the authors themselves considered to be exploratory. The purpose of the present study was to produce an updated meta-analysis including only data from the primary analyses of clinical trials evaluating the role of PFO closure in the secondary prevention of recurrent stroke, since several texts previously published contained data from both original clinical trials and a follow up study. Methods Search strategy The study started with a search onMedline (PubMed) database, using the query “patent foramen ovale” AND “stroke” AND “closure” with the filter “clinical trial”. The search took place on July 2018, and no articles were excluded based on publication date. The search yielded 40 articles. A further search was carried out in a second database, ISI Web of Knowledge, using the same query, with the filter “article”, on December 2018, yielding 840 articles (Figure 1, supplementary file 1). Additional studies were found after searching the references of previous review articles and other relevant sources, including articles related to the topic in question as well as articles citing the selected articles. Inclusion criteria Only human studies were included, and only interventional studies comparing PFO closure with medical therapy were considered within the scope of this review. Exclusion criteria The following were excluded: mechanistic studies, animal studies, studies of PFO physiology, case reports, editorials, review papers, study protocols, non-randomized studies, duplicate studies (if found), systematic reviews and/or meta-analyses, sub-group analyses of included studies, follow-up data of included studies, cost analyses or surveys, comparison between medical treatments, comparison between closure devices, studies on PFO closure only, guidelines, genetic and pathological studies. Statistical analysis We aimed at presenting an overview of clinical trials evaluating interventional studies comparing PFO closure with medical therapy. The meta-analysis was carried out by using the Comprehensive Meta-analysis Software V.2.0 (Biostat, New Jersey, USA). Random- effects analyses were carried out, given the considerable heterogeneity of some of the data. The parameters chosen for analysis and also for the meta-analyses were: stroke, transient ischemic attack and atrial fibrillation. Risk ratios were calculated. A level of significance of 5% was used. Results were reviewed by a biostatistician (CS). Quality assessment of studies and data extraction Study quality and eligibility were independently assessed by two researchers. Different opinions regarding the relevance of articles were solved by consensus between the authors. Global article quality assessment was carried out according to the method used by Haffar and colleagues (supplementary file 2). 20 Results A total of six articles were identified and selected for further study. 3,5,6,8-10 Interobserver agreement was 100%. Between 2012 and 2018, six randomized controlled trials (RCTs) comparing closure of PFO with medical